ES2918773T3 - Methods and compositions related to neurodegenerative diseases - Google Patents

Abstract

The present invention provides a method of diagnosing or evaluating a neurodegenerative disease in a test subject, comprising: (i) providing a sample containing protein that has been obtained from the test subject; (ii) determining the concentration, amount or extent of expression of at least one specific protein isoform and/or glycoform derived from a protein biomarker selected from the group consisting of: clusterin precursor; Apolipoprotein A-IV precursor; Apolipoprotein C-III precursor; transthyretin; galectin 7; C4 precursor complement; alpha-2-macroglobulin precursor; IG alpha-1 chain C; histone 2b; Ig Lambda Chain C Region; fibrinogen gamma chain precursor; complement factor h; H4 heavy chain precursor between alpha-trypsin; C3 precursor complement; gamma or beta actin; haptoglobin precursor; and the serum albumin precursor, or a fragment thereof; (iii) Comparison of said concentration, amount or degree determined in (ii) with a reference of a control subject with a specific neurodegenerative disease, dementia or disease stage, or of a control subject not having a neurogenerative disease or dementia ; and (iv) based on the level of at least one specific protein isoform and/or protein biomarker glycosformation in the test subject relative to the control, making a diagnosis or assessment as to the presence and/or stage of the neurodegenerative disease or dementia of the test subject. Related products and systems for use in said method are also provided. (Automatic translation with Google Translate, without legal value)

Description

DESCRIPTION

Methods and compositions related to neurodegenerative diseases

field of invention

The present invention relates to methods related to neurodegenerative diseases, including Alzheimer's disease. Specifically, the present invention identifies and describes protein isoforms that are differentially expressed in the Alzheimer's disease state relative to their expression in the normal state, and, in particular, identifies and describes proteins associated with Alzheimer's disease.

Background of the invention

Dementia is one of the main public health problems of the elderly and in our aging populations, the increasing number of patients with dementia is imposing a significant financial burden on health systems worldwide. More than half of dementia patients have Alzheimer's disease (AD). The prevalence and incidence of AD have been shown to increase exponentially. The prevalence for AD in Europe is 0.3% for ages 60-69 years, 3.2% for ages 70-79 years, and 10.8% for ages 80-89 years. Hofman et al. 1991). Survival time after the onset of AD is approximately 5 to 12 years (Friedland 1993).

Alzheimer's disease (AD), the most common cause of dementia in older individuals, is a debilitating neurodegenerative disease for which there is currently no cure. It destroys neurons in parts of the brain, primarily the hippocampus, which is a region involved in encoding memories. Alzheimer's disease leads to an irreversible progressive loss of cognitive functions and functional autonomy. Early signs of AD can be mistaken for simple forgetfulness, but in those ultimately diagnosed with the disease, these early signs progress inexorably to more severe symptoms of mental decline. While the time it takes to develop will vary from person to person, advanced signs include severe memory impairment, confusion, language disturbances, personality and behavior changes, and impaired judgment. People with AD can become uncommunicative and hostile. As the disease progresses, its course ends in profound dementia, patients are unable to care for themselves and often require institutionalization or professional care at home. While some patients may live for years after being diagnosed with AD, the average life expectancy after diagnosis is eight years.

In the past, AD could only be definitively diagnosed by brain biopsy or autopsy after a patient died. These methods, which demonstrate the presence of the characteristic plaque and tangle-like lesions in the brain, are still considered the gold standard for pathological diagnoses of AD. However, in the clinical setting, a brain biopsy is rarely performed, and diagnosis depends on a range of neurological, psychometric, and biochemical tests, including measurement of biochemical markers, such as ApoE and tau proteins or beta-amyloid peptide in the brain. cerebrospinal fluid and blood.

Biomarkers possibly hold the key to the next step in diagnosing AD and other dementias. A biological marker that meets the requirements for diagnostic testing for AD would have several advantages. An ideal biomarker would be one that identifies AD cases very early in the disease, before degeneration is seen on brain imaging and neuropathology. A biomarker could be the first indicator to start treatment as soon as possible and also very valuable to detect the effectiveness of new therapies, particularly those that focus on preventing the development of neuropathological changes. Repetitive measurement of the biological markers of the invention would also be useful in following the development and progression of the disease.

Markers related to pathological features of AD; plates and tangles (AI3 and tau) have been the most extensively studied features. The most promising have been CSF concentration studies of AI3(1-40), AI3(1-42) and tau or the combination of both proteins in AD. Many studies have reported a decrease in AI3(1-42) in lCr, while the total AI3 or AI3(1-40) protein concentration remains unchanged (Ida, Hartmann et al. 1996; Kanai, Matsubara et al. 1998 ; Andreasen, Hesse et al. 1999).

Recognizing that CSF is a less desirable sample and that "classic" markers of AD pathology, including amyloid and tau, are not reliably detectable in other fluids, several efforts have been made to identify markers of AD pathology. additional proteins in blood and blood products such as serum and plasma. A group of blood proteins that are differentially expressed in the AD state, relative to their expression in the normal state, is described in WO2006/035237 and includes the protein clusterin, which has been previously associated with AD pathology. in the brain of affected individuals. The value of clusterin as a potential biomarker in AD has been explored by various groups, both in cerebrospinal fluid (CSF) and blood, often with conflicting results. One possible explanation for the discrepancy between CSF and brain clusterin levels is the effect of protein glycosylation, which may serve to mask epitopes recognized by antibodies used in immunoassays to measure clusterin. In fact, Nilselid et al. (2006) demonstrated that accurate quantification of clusterin in human CSF was only possible when all glycan residues were enzymatically removed from clusterin, prior to measurement by ELISA. In their study, they found that the amount of clusterin measured by two specific antibodies against clusterin alpha and beta chains increased by about 70% after deglycosylation. Importantly, clusterin levels were generally elevated in male AD patients, relative to healthy male controls, their study failed to show the diagnostic utility of measuring CSF levels of naturally glycosylated clusterin levels or those of of ex vivo deglycosylated protein. Furthermore, they found no difference in clusterin levels between women with AD and the female control group. The authors conclude that there was no overall difference in clusterin glycosylation levels between EA and control groups, but contradict this by suggesting that protein microheterogeneity (glycosylation, phosphorylation, etc.) might be another useful target in diagnosis or prognostic monitoring of the disease.

In Maja Amedjkouh Puchades: "Development of proteomic methods for studying cerebrospinal fluid proteins involved in Alzheimer 's disease.", January 1, 2003 (2003-01-01), pages 1-68, two clusterin isoforms are reported to be present in CSF, which are altered in AD patients versus normal controls.

Brief description of the invention

In light of this uncertain technique and desiring to develop a minimally invasive diagnostic test using blood instead of CSF, the inventors have surprisingly shown that clusterin glycosylation in human plasma is highly heterogeneous, with more than 40 different isoforms identified to date. date. Furthermore, a small subset of only 8 of the identified glycoforms is consistently upregulated among AD patients and those with mild cognitive impairment. Furthermore, the levels of these same glycoforms can predict the severity and rate of progression of AD within an individual.

The present inventors have previously determined several plasma biomarkers for Alzheimer's disease (see US7,897,361). However, they found that selected immunoassays and reaction monitoring experiments did not fully replicate the results they obtained for the same biomarkers, using 2-dimensional gel electrophoresis (2DE). The inventors investigated whether this difference could be due to specific post-translational events that were not being replicated in the validation experiments.

The inventors surprisingly found that post-translational events created distinct isoforms of the protein, eg glycoforms, which were differentially expressed in different forms and stages of dementia. Consequently, the inventors have identified more powerful biomarkers for dementia and, as a result, may provide more sophisticated methods for the diagnosis, prognosis, and monitoring of dementia, such as Alzheimer's disease.

In general terms, the present invention relates to a method for evaluating the level of hippocampal atrophy in a test subject with Alzheimer's disease (AD).

A protein in vivo can be present in several different forms. These different shapes can be produced by alternative splicing; by alterations between alleles, for example, single nucleotide polymorphisms (SNPs); or it may be the result of post-translational events such as glycosylation (glycoforms). A glycoform is an isoform of a protein that differs only with respect to the number or type of glycan attachment.

The present invention provides a method for evaluating the level of hippocampal atrophy in a test subject with Alzheimer's disease (AD), comprising:

(i) providing a protein-containing sample that has been obtained from the test subject;

(ii) determining the concentration, amount or degree of expression of at least one glycoform of clusterin, where the antennal tetra-glycoforms of the at least one glycoform of clusterin, as a proportion of the total of all glycoforms of the same glycoprotein fragment , is quantified;

(iii) comparing said concentration, amount or degree determined in (ii) with a reference of a control subject having eA with a low level of hippocampal atrophy;

(iv) based on the level of the at least one glycoform of clusterin in the test subject relative to control, making an assessment of the level of hippocampal atrophy in the test subject,

wherein, a lower relative level of tetra-antennary glycoforms in the test subject sample, compared to the relative level of tetra-antennary glycoforms in the control subject reference, indicates that the test subject is expected to have a relatively higher level of hippocampal atrophy;

wherein said at least one glycoform of clusterin comprises a glycosylated fragment of human clusterin consisting of HN*STGCLR (SEQ ID NO: 2), wherein "N*" indicates the glycan attachment residue; Y

wherein, the protein-containing sample is selected from the group consisting of: blood plasma, blood cells, or serum.

Said at least one glycoform of clusterin may further comprise:

a) any of the clusterin glycopeptides set forth in Table 3A, Table 3B, Table 3C, Table 5, Table 6 and/or Table 7; me

b) an I364N glycan selected from the group consisting of:

IJ64N_SA1-(HexNAc-Hex)2-core; G64N_SA2-(HexNAc-Hex)2-middle; G64N_SA1-(HexNAc-Hex)3-middle; G64N_SA2-(HexNAc-Hex)3-middle; G64N_SA1-(HexNAc-Hex)4-middle; G64N_SA3-(HexNAc-Hex)3-middle; G64N_SA2-(HexNAc-Hex)4-middle; and G64N_SA3-(HexNAc-Hex)4-core.

The skilled person will be aware that there are a variety of techniques suitable for measuring the amount or concentration of specific glycoforms. This includes the use of non-human antibodies generated by immunization with specific isoforms of the proteins of this invention, where such antibodies have the required specificity for the diagnostic isoform, particularly glycoforms. In particular, the use of synthetic peptides of Sequence IDs 2-10 with the appropriate glycan structures. Such peptides are not found in nature and therefore must be prepared ex vivo through natural digestion of clusterin or by use of synthetic chemistry in vitro.

More specifically, methods including measurement by gel electrophoresis or LC-MS/MS are contemplated herein.

In some cases, the relative amount of each glycoform is calculated by comparison with an equivalent reference glycoform, labeled with heavy isotopes, using Selected Reaction Monitoring mass spectrometry. In particular, the heavy isotope-labeled reference glycoform may be a synthetic glycopeptide in which one or more heavy isotopes of H, C, N or O are substituted within the peptide or sugar components of said glycoform.

In some cases, the heavy isotope-labeled reference glycoform is an enriched natural glycoform that has been labeled with an isotopic mass tag, wherein said isotopic mass tag with one or more heavy isotopes of H, C, N or O and wherein said mass tag is capable of reacting with the peptide or sugar components of said glycoform.

In some cases, the relative amount of each glycoform is calculated by comparison with an equivalent glycoform labeled with an isobaric mass tag, as generally disclosed in European Patent 2,115,475, where:

(i) each sample taken from the test subject is labeled with a member of a set of isobaric mass labels, to create a labeled analytical sample;

(ii) a standard reference panel of enriched glycoforms is separated into two to six aliquots and each aliquot is separately labeled with additional members of the same set of isobaric mass labels as the labeled analytical sample and each aliquot is independently labeled, from the reference panel, is mixed in a predefined ratio to create a clinically relevant concentration curve, such as a standard reference mixture; (iii) an equal volume of the labeled analytical sample and the standard reference mix are mixed to form the MS calibrator sample; Y

(iv) The MS calibrator sample prepared in step (iii) is analyzed by mass spectrometry. In particular, the set of isobaric mass tags may be a set of Tandem Mass Tags.

In certain instances, in accordance with the present invention, glycoforms are measured using summation-scale Selected Reaction Monitoring (SRM) mass spectrometry.

In certain cases according to the present invention, the glycoforms are not labeled.

In certain instances according to the present invention, the method does not comprise subjecting the sample to gel electrophoretic separation and/or does not comprise subjecting the sample to enrichment by immunoprecipitation.

In certain instances according to the present invention, the glycoforms are glycoforms and are measured by a method essentially as described in Example 6.

In some cases, according to the present invention, the at least one glycoform can be measured by immunological assay, such as Western Blot or ELISA.

Measurement of glycan structures on clusterin can be done by various methods. In two-dimensional gel electrophoresis, the addition or removal of sugar groups within the glycan backbone will affect both the apparent molecular mass and isoelectric focus point of the clusterin, generating a "train" of spots within the gel. Said stain trains are well known to those skilled in the art. As an example, a plasma protein from a subject suspected of having dementia is subjected to two-dimensional gel electrophoresis. After completion of the second dimension, the gel is stained with a selective protein or sugar stain to reveal individual protein spots or glycoprotein spots, respectively. Typically, an image of the entire gel is captured with a CCD camera, and the relative abundance of each spot is calculated as a function of staining intensity using commercially available software such as SameSpots (Non-Linear Dynamics, UK). The stain train comprising clusterin isoforms can be identified by comparison with a reference gel. Alternatively, the spots can be cut out of the gel and the proteins identified by mass spectrometry. Ultimately, the relative abundance of each spot representing the different clusterin isoforms is determined.

This invention also provides a method for stratifying a plurality of test subjects according to their stage of Alzheimer's disease (AD), comprising:

carrying out the method according to the present invention on at least one test sample from each of the human test subjects; Y

based on the level of the at least one specific clusterin glycoform in each of the test subjects, stratifying the test subjects into a more or less advanced stage of AD; wherein test subjects are classified according to their predicted degree of hippocampal atrophy.

This invention also provides a method for determining the efficacy of a treatment for Alzheimer's disease (AD) in a mammalian subject, comprising:

determine the level of one or more glycoforms of clusterin by carrying out the method according to the present invention, prior to treatment of the subject and at least once during or after treatment of the subject, wherein successful treatment is demonstrated, by the level of said or more glycoforms of clusterin, which remains stable or returns to more normal levels, wherein, the subject is human.

The invention will now be described in more detail, by way of example and without limitation, with reference to the accompanying drawings.

Brief description of the figures

Figure 1. Theoretical and actual distributions of clusterin glycoform in 2DE. Panel A - Mathematical modeling of the alpha and beta chain of clusterin. Additional series for tetra-antennary structures can be modeled in a similar way (not shown). Panel B - 2DE spots of immunoprecipitated protein. Sixteen different spots were analyzed and fully sialylated N-glycans were the most abundant structure at each glycosylation site in all 16 spots. The change in pI observed for different gel spots is likely due to glycosylation through alterations in the number of antennae and sialic acids.

Figure 2. Tabular representation of individual glycoforms of four clusterin peptides detected in 16 spots on 2DE gels (Peptide A: SEQ ID NO: 2; Peptide B: SEQ ID NO: 11; Peptide C: SEQ ID NO: 7; Peptide D : SEQ ID NO: 10).

Figure 3. Vector diagram illustrating the shift in 2DE coordinates associated with removal of specific glycan units from N-linked carbohydrates.

Figure 4. Structure of the substrate NA3 (Dextra-UK Ltd, Catalog No: C1124) Molecular Weight = 2006.82 Da Chemical Formula: C76H127N5O56.

Figure 5. ESI-TOF spectrum of intact NA3 substrate showing the presence of doubly charged molecular ions at m/z 1003.87 and related sodium and potassium cations.

Figure 6. MS/MS spectrum of m/z 1050.74 of [M+3H]3+ molecular ion for clusterin glycopeptide of Molecular Weight 3149.22 Da. The fragment ions allow the structure of the glycan to be deduced with the fragment ion at m/z 574.47 representing the [Peptide+HexNac]+ moiety. Thus, the "naked" Peptide sequence is HN*STGCLR (SEQ ID NO: 2) and a fully sialylated bi-antennary glycan structure, SA2-(HexNAc-Hex)2, is attached to the asparagine residue (N *).

Figure 7. Mass spectrum showing the molecular ions of two sialylated forms of the tetraantennary glycopeptides HN*STGCLR (SEQ ID NO: 2) observed in instances of low atrophy and not observed in high atrophy.

Figure 8. Relative percentage of tetra-antennary glycoforms within eight individuals with low and high levels of hippocampal atrophy.

Figure 9. Shows box plots of significantly regulated I364N clusterin glycopeptides from the Discovery Cohort (Orbitrap Fusion) A) l364N_SA1-(HexNAc-Hex)2-core; B) l364N_SA2-(HexNAc-Hex)2-central; C) l364N_SA1-(HexNAc-Hex)3-central; D) l364N_SA2-(HexNAc-Hex)3-central; E) l364N_SA3-(HexNAc-Hex)3-central; and F) l3B4N_SA3-(HexNAc-Hex)4-central.

Figure 10. Shows box plots of significantly regulated I364N clusterin glycopeptides from the Replication Cohort (Orbitrap Fusion) A) l364N_SA1-(HexNAc-Hex)2-core; B) l364N_SA1-(HexNAc-Hex)3-central; and C) 1364N_SA2-(HexNAc-Hex)3-central.

Figure 11. Shows box plots of significantly regulated clusterin I364N glycopeptides from Cohorts of Combined Replication and Discovery (Orbitrap Fusion) A) l364N_SA1-(HexNAc-Hex)2-central; B) I364N_SA2-(HexNAc-Hex)2-central; C) l364N_SA1-(HexNAc-Hex)3-central; and D) l364N_SA2-(HexNAc-Hex)3-central.

Figure 12 shows box plots of significantly regulated I364N clusterin glycopeptides from Discovery Cohort by SRM analysis (TSQ Vantage) A) l364N_SA1-(HexNAc-Hex)2-core; B) l364N_SA2-(HexNAc-Hex)2-central; C) l364N_SA1-(HexNAc-Hex)3-central; D) l364N_SA2-(HexNAc-Hex)3-central; and E) l364N_SA1-(HexNAc-Hex)4-central.

Figure 13 shows an SDS-PAGE image of normal human plasma without albumin/IgG. Red bars represent breakpoints and numbers represent the band number used for Orbitrap analysis to identify clusterin glycopeptides.

Figure 14 shows the total ion chromatogram (TIC) of band #4 - #9 of plasma depleted using the glyco-SRM method. Eight glycopeptides of clusterin I364n (m/z 953.71, 1050.74, 1075.42, 1172.45, 1197.13, 1269.49, 1294.17, 1391.53) and the fragment ions were used as precursors. at m/z 366.14, 574.56 and 657.24 were established as transition ions for each precursor.

Figure 15. XIC sample of band #7 showing various retention times and peak areas of eight glycoforms at the I364N site.

Figure 16 . Shows box plots of significantly regulated I364N clusterin glycopeptides from combined Discovery and Validation Cohorts by SRM analysis (TSQ Vantage) A) l364N_SA1-(HexNAc-Hex)2-core; B) 1364N_SA2-(HexNAc- Hex)2-central; C) l364N_SA1-(HexNAc-Hex)3-central; D) l364N_SA2-(HexNAc-Hex)3-central; E) 1364N_SA1 -(HexNAc-Hex)4-central; F) l364N_SA3-(HexNAc-Hex)3-central; G) l364N_SA2-(HexNAc-Hex)4-central; and H) l364N_SA3-(HexNAc-Hex)4-central.

Figure 17 shows Table 1A - The Clusterin GlycoMod v1.0 database (control plasma) (SEQ ID NOs: 2-7 and 10, respectively).

Figure 18 shows Table 1B - The Clusterin GlycoMod v1.1 (plasma DCL/EA) database (SEQ ID NOs: 2-7, 9, 10, and 8 respectively).

Detailed description

In describing the present invention, the following terms will be used and are intended to be defined as indicated below.

The term "subject" includes a human or an animal. In accordance with certain embodiments of the present invention, the subject may have been previously diagnosed with AD and/or previously diagnosed with mild cognitive impairment (MCI). The subject is preferably a human. The subject may be a human being at least 60 years of age, optionally at least 70 or at least 80 years of age.

The term "diagnosis" as used herein includes the provision of any information regarding the existence, nonexistence, or likelihood of the disorder in a patient. It further includes the provision of information about the type or classification of the disorder or the symptoms that are experienced or may be experienced in relation to it. This may include, for example, diagnosis of the severity of the disorder. It covers the prognosis of the medical course of the disorder, for example, its duration, severity, and the course of progression, from MCI to Alzheimer's disease.

Currently, disease status is assessed by disease duration from onset to present (longer duration equals more severe disease) and clinical assessment measures. These assessment measures include clinical tests of memory and other cognitions, clinical tests of function (daily living skills), and clinical assessments of global severity. Trials of potential therapies in AD are currently being evaluated against these measures. The FDA and other drug approval agencies require, as part of these evaluations, measures of both cognition and global function. The Global Dementia Scale is one such measure of global function. It is assessed by post-assessment of severity, including cognition and function, against a standardized set of severity criteria.

The term "alleviating" as used herein in relation to Alzheimer's disease means any way of reducing one or more symptoms or undesirable effects thereof. Any improvement in the patient's Alzheimer's disease falls under the term "relief." Improvement may also include slowing the progression of the disease.

As used herein, "assessing" AD includes providing information regarding the type or classification of the disease or the symptoms that are or may be experienced in connection therewith. This specifically includes the prognosis of the medical course of the disease, eg, its duration, severity, and the course and rate of progression from, eg, MCI or presymptomatic AD to clinical AD. This also includes the forecast of brain pathology associated with AD, such as fibrillary amyloid burden, cortical and hippocampal atrophy, and accumulation of neurofibrillary tangles. The evaluation may be of an aggressive form of AE and/or a poor prognosis.

As used herein, "biological sample" refers to any biological fluid, cell or tissue sample isolated or obtained from the subject. In accordance with the present invention, the "protein-containing sample" can be any biological sample as defined herein. The biological sample may, in certain cases, comprise blood plasma, blood cells, serum, saliva, urine, cerebrospinal fluid (CSF), or a tissue biopsy. The biological sample may have been stored (eg, frozen) and/or processed (eg, to remove cellular debris or contaminants) prior to determining the amount (eg, concentration) of at least one protein isoform and/or glycoform. in question found in the sample.

Exemplary Clusterin Glycoform Analysis

Clusterin (Apolipoprotein J; SP-40,40; TRPM-2; SGP-2; pADHC-9; CLJ; T64; GP III; XIP8) is a highly conserved disulfide-linked secreted heterodimeric glycoprotein of 75-80 kDa, but truncated forms targeting the nucleus have also been identified. The protein is constitutively secreted by various cell types, including epithelial and neuronal cells, and is an important protein in physiological fluids, including plasma, milk, urine, cerebrospinal fluid, and semen.

Preferably, the clusterin comprises or consists of an amino acid sequence that has at least 70%, 80%, 90%, 95%, 99%, or 100% identity to the human clusterin sequence described in UniProt Accession No.: P10909, sequence version 1 and GI No. 116533 (SEQ ID NO: 1), calculated on the total length of said human clusterin sequence; or a fragment thereof, comprising at least 5, 10, 15, 20, 30, 50, 100, 150, 200, 250, 300, 350, 400, 425, or 449 contiguous amino acids.

Clusterin gene expression is significantly elevated in Alzheimer's disease (AD) brain (May et al., 1990) and plasma clusterin levels have also been shown to correlate with AD progression (Thambisetty et al., 2010). The inventors have previously identified several plasma clusterin isoforms as candidate biomarkers for AD, using two-dimensional gel electrophoresis (2DE).

However, the use of immunoassays and unmodified peptides in selected reaction monitoring (SRM) experiments did not fully replicate the regulation observed in 2D ^e . The inventors hypothesized that this disconnection might be due to alterations in specific post-translational events that were not replicated in validation studies. Clusterin is a highly glycosylated secreted protein and because glycosylation plays an important role in the physiological functions of clusterin (Stuart et al., 2007) the inventors proposed that detailed profiling of plasma clusterin and comparison of glycosylation profiles observed in different clinically classified subjects, eg, patients with low or high hippocampal atrophy, may reveal more potent biomarker isoforms.

Guided by the observations regarding clusterin glycoforms, which demonstrate 13-N-acetylglucosamin-idase activity in plasma, the inventors also devised a new assay using a defined substrate to measure this specific activity.

Example 1 . Plasma gel electrophoresis analysis of clusterin isoforms

Methods

Human clusterin was enriched by immunoprecipitation (IP) from albumin/IgG depleted plasma, using an anti-clusterin monoclonal antibody (Millipore). Immunoprecipitated proteins were first analyzed by Western blot as quality control and then separated by two-dimensional electrophoresis (2DE) or SDS-PAGE. Spots and the single band (#3) of interest were excised, reduced, alkylated and digested in gel with trypsin before analysis by mass spectrometry (MS). Samples were analyzed via LC-MS/MS using nanoflow reverse-phase chromatography (EASY-nLC II, ThermoFisher Scientific) and a Top20 collision-induced dissociation (CID) method (Orbitrap Velos, ThermoFisher Scientific). Glycopeptides were manually identified by the presence of glycan-specific oxonium ion fragments, m/z 204.08 for N-acetylhexosamine, [HexNAc]+, m/z 366.14 for hexose-N-acetylhexosamine, [Hex-HexNAc ]+ and m/z 657.24 for N-acetylneuraminic acid-hexose-N-acetylhexosamine [NeuAc-Hex-HexNAc]+ in MS/MS spectra.

Results

2DE spots

Initially, a mathematical model was used to create an artificial map of the various clusterin glycoforms for the separate alpha and beta chains (Figure 1A). Further refinement of this approach was used to classify individual clusterin-related glycoforms using simple (x,y) coordinates. In this way, the inventors were able to predict the content of individual 2DE spots and show that multiple glycoforms share discrete coordinates. Therefore, 2DE spots are likely to be composite mixtures containing various glycoforms. This information was then used to aid in the interpretation of complex LC/MS/MS data, which subsequently provided some pretty useful insights.

First, it became clear that the main components within each of the 2DE spots were, without exception, always completely sialylated forms, being tetra, tri or biantennary structures (Figure 1B). This was surprising because differences in sialic acid had originally been predicted to be the main cause of separation of different shapes during electrophoresis, with loss of 291Da concurrent with decreasing charge, resulting in a shift towards a lower isoelectric point. essential. However, LC/MS/MS results (Figure 2) indicated a trend towards fewer antennae numbers and this suggested successive removal of entire antennae as a more pronounced effect on 2DE spot location. A basic vector was devised to illustrate this phenomenon (Figure 3) and it turns out that the detected clusterin glycoforms now indicate evidence to support both removal of sialic acids alone, and removal of entire antennae suggesting distinct neuraminidase and 3-β activities. N-acetyl-glucosaminidase, respectively.

Example 2 - Design of a substrate assay to measure the specific activity of n-acetyl-glucosaminidase in plasma

Results of glycan analysis of 16 different clusterin isoforms visible on two-dimensional gel electrophoresis showed sequential removal of sialic acids and complete antennas. Several of the truncated glycoforms appeared to correlate with clusterin protein spots previously identified as candidate biomarkers of AD and MCI. Until now, a detailed analysis of the glycosylation of these clusterin isoforms has not been performed and it was surprising to discover that most of the disease-associated modifications in plasma clusterin could be explained by the activity of a single glycosidase, viz. 3-N-acetyl-glucosaminidase.

The inventors thus established a specific assay method for determining 3-N-acetylglucosaminidase activity in tissue or body fluid samples taken from subjects suspected of having or previously diagnosed with MCI, AD or other dementia. The artificial NA3 glycan substrate (Figure 4) contains 131.2 and 131.4 linkages between contiguous 13-N-acetyl-glucosamine and mannose subunits and is a preferred substrate for differentiating 31.2 and 31.4 N-activity. acetyl-glucosamine-idase in plasma. The molecular weight of the substrate is 2006.82 Da and a [M+2H]2+ ion was detected at m/z 1003.8, using an ESI-TOF mass spectrometer (FIG. 6).

The NA3 substrate is added to an appropriate sample of tissue or body fluid from a subject with suspected or previously diagnosed dementia, to achieve a final concentration of 300 - 1,000 pg/|jl and incubated at 37°C for 4-24 hours. The test sample is then centrifuged to remove debris and an aliquot is sent for LC-MS/MS analysis. Measurement of molecular ions corresponding to the loss of two or one antenna indicates 31.2 and 31.4 activity of N-acetyl-glucosaminidase respectively.

Example 3 - Analysis of Immunoprecipitated Clusterin to Create a Unique Glycopeptide Reference Resource for Clusterin:_The Clusterin GlycoMod v1.0 Database

A representative pooled clinical plasma sample was used to develop the methodology and assemble an "observation-based" database containing 41 distinct glycoforms associated with anticipated glycosylation consensus sites, within the amino acid sequence. For each glycopeptide, the charge state m/z and the retention time (RT) of the analyte were tabulated (see Table 1A). Unequivocal annotation of the glycopeptide required detection of the [Peptide+HexNAc]+ fragment ion in the corresponding MS/MS spectra and interpretation of additional fragment ions, related to the sequential dissociation of individual glycan subunits. An example MS/MS spectrum is shown (Figure 7) and the current iteration of the Clusterin GlycoMod database is provided ⁱⁿ Table 1a. An updated iteration of the Clusterin GlycoMod database is provided in Table 1B.

Using immunoprecipitation and LC/MS/MS, 41 glycopeptides spanning 5 of the 6 anticipated N-linked glycosylation consensus sites on plasma clusterin have been characterized. In total 41 different N-linked glycopeptides have been characterized and are listed here. The distribution of glycans at these 5 sites was consistent with a CV of <15% (n=3 from two plasma samples) indicating the technical and biological reproducibility of the method.

The inventors have previously demonstrated 5 of 6 predicted N-linked glycosylation sites within human plasma clusterin (GlycoMod v1.0 database). The skilled artisan would understand that expansion of the GlycoMod database to cover all N-linked and O-linked sites of all protein biomarkers is within the scope of the present invention. Indeed, the inventors have subsequently completed the mapping of the sixth N-linked site in human clusterin, as set forth in Table 1B (FIG. 18), showing clusterin glycopeptides from version 1.1 of the Clusterin GlycoMod database ( plasma MCI/EA).

Example 4 - Immunoprecipitated Clusterin Assay to Compare Low and High Atrophy Individuals

The inventors have identified certain differentially regulated isoforms of clusterin in the plasma of AD patients, relative to controls without dementia. In addition, certain clusterin-comprising stains on 2DE gels have also been shown to correlate with the level of hippocampal atrophy, while other isoforms correlate with the subsequent rate of disease progression in AD.

The inventors obtained plasma samples from four previously diagnosed AD subjects who had a low level of hippocampal atrophy and from four previously diagnosed AD subjects with high hippocampal atrophy. Clusterin was enriched by immunoprecipitation and subjected to the LC-MS/MS method described above. Surprisingly, they identified that the extent of glycan pruning correlated with hippocampal atrophy. In patients with low levels of hippocampal atrophy there was little evidence of plasma clusterin pruning. In contrast, plasma clusterin from subjects with high levels of hippocampal atrophy was generally pruned to remove one or more complete antennas within N-linked glycans.

As an example, two sialylated forms of the tetra-antennary glycopeptide HN*STGCLR (SEQ ID NO: 2) are observed as triply charged molecular ions at m/z 1391 and m/z 1294.17, but only in individuals with low atrophy (Figure 7 ). These remnants are therefore potential surrogate biomarkers consistent with the degree of hippocampal atrophy.

Using data from all N-linked glycans monitored by the LC-MS/MS method, the inventors observed a consistent reduction in the level of tetra-antennary glycans in subjects with high levels of hippocampal atrophy, compared to those with low levels. . Based on the total glycoform signal for the N-linked glycosylation site in the tryptic peptide HN*STGCLR (SEQ ID NO: 2) of the clusterin beta chain (FIG. 8).

Example 5 - Immunoprecipitated Clusterin Validation Assay to Compare Low and High Atrophy Individuals

Having identified that changes in clusterin glycosylation patterns correlate with the degree of atrophy within a small cohort of clinical samples, a further validation study was performed in an additional cohort of Alzheimer's disease patients with known levels of atrophy. of the hippocampus. Additional bioinformatic approaches were also evaluated for their impact on glycoform-based class segregation profiles and a new, higher-sensitivity mass spectrometer was employed with the expectation of identifying additional diagnostic clusterin glycoforms. To ensure correlation with previous data, samples from the original 4 x 4 Cohort (Discovery Cohort) used in Example 4, were re-analyzed using the new methods along with a separate cohort of 20 new EA samples ( n=10) and matched controls (n=10)(Replication Cohort). All sample details are provided in Table 2.

Cohort sample details

Table 2. Sample details associated with cohorts 4 vs 4 10 vs 10

Methods

Clusterin was enriched from each sample, as described above. The relevant protein band was excised, reduced, alkylated and digested with trypsin. After cleanup, the clusterin digests were divided into two aliquots and each analyzed by nanoflow high performance liquid chromatography and Orbitrap Velos Pro or ultrahigh performance liquid chromatography systems and Orbitrap Fusion Tribrid LC-MS/MS (all Thermo Scientific equipment, Hemel Hempstead, UK). The data were apparently similar but, as expected, more glycosylated clusterin peptides were identified in Fusion, so all further analyzes were performed on the Fusion dataset.

Bioinformatics

Raw mass spectrometer data was processed using Proteome Discoverer software (Thermo Scientific). The ion intensities for the glycosylated clusterin peptides and their fragments described in Tables 1A and 1B were exported to an Excel spreadsheet (Microsoft Corp). A summation scaling technique was used to normalize the data and calculate significance (p) values for each glycopeptide, comparing the mean values between the low and high atrophy groups in the Discovery Cohort, Replication Cohort, and in a combined analysis of both Cohorts as a single group. Student's t-test was used to identify peptide-associated glycoforms that switch significantly between high and low atrophy, resulting in one-tailed p-values for each glycopeptide (see Tables 3A, 3B and 3C).

Results

Using the IP-LC/MS/MS workflow on the Orbitrap Fusion Tribrid, coverage could be extended to all six known clusterin N-glycosylation sites: α64N, α81N, α123N, I364N, I3127N, and I3147N. By monitoring the specific glycan fragments, it was also possible to assign various antennal structures at the six sites and perform relative quantification based on total ion counts. In total 42 different glycan structures were detected. While most glycosylation sites showed no regulation of glycan structures between high and low levels of hippocampal atrophy, two sites - I364N and I3147N - showed significant regulation between clinical groups. Specific glycan structures showing (p < 0.05) changes between clinical groups in the combined Discovery, Replication and Cohort analyses, are indicated in Tables 3A, 3B and 3C respectively. Box plots were created for each glycopeptide to illustrate the separation achieved between the two groups (Figures 9-11).

Interestingly, six glycoforms at the I364N HN*STGCLR (SEQ ID NO: 2) glycosylation site were found to be significantly decreased in the 4 high atrophy (Alzheimer's) samples compared to the 4 low atrophy (mild cognitive impairment) samples from the Discovery Cohort, when measured on the Orbitrap Fusion. This included the observed tetra-antennary glycopeptide sialylated forms as triply charged molecular ions at m/z 1391.54 which was consistent with previous Velos data analysis, confirming the robustness of this glycoform as a diagnostic marker for differentiate mild cognitive impairment from Alzheimer's disease, when measured on a different LC-MS/MS platform.

In the largest Replication Cohort, three I364N glycopeptide glycoforms were significantly reduced in the high atrophy samples. These include the glycoforms SA1-(HexNAc-Hex)2, SA1-(HexNAc-Hex)3 and SA2-(HexNAc-Hex)3 seen at m/z 953.71, 1075.42, 1172.45 in the spectra. As all of these glycoforms were also reduced in patients with high atrophy in the Discovery Cohort, this further supports their utility as prognostic biomarkers in patients with confirmed Alzheimer's disease.

When the results of the two cohorts were combined, we again saw that changes in glycoforms found at the I364N site correlated with atrophy, with four glycoforms significantly reduced over high atrophy, for example, SA1-(HexNAc-Hex)2 , SA2-(HexNAc-Hex)2, SA1-(HexNAc-Hex)3 and SA2-(HexNAc-Hex)3 at m/z 953.71, 1050.74, 1075.42 and 1172.45 respectively.

T l A m i i nifi iv n l li i l rin 4 v 4)

T l B m i i nifi iv n l li i l rin v 1 )

Table 3 m i i nifi iv n l li i ^ l rin m in 13 vs 14)

conclusion

The use of Orbitrap Fusion increased the total glycoform coverage of 4 to 6 N-binding sites. Several I364N site glycoforms are significantly reduced in the plasma of patients with Alzheimer's disease, compared to individuals with mild cognitive impairment. Four of these glycoforms are also reduced in Alzheimer's patients with high levels of hippocampal atrophy. In combination, this confirms the utility of clusterin isoforms as diagnostic and prognostic markers for Alzheimer's disease.

Example 6 - Development and preliminary testing of a Selective Reaction Monitoring (SRM) method for 8 glycoforms of clusterin in human plasma

In preparation for higher throughput measurements within a much larger number of clinical samples, a specific Selective Reaction Monitoring (SRM) method was also developed to measure clusterin-specific glycopeptides. This recently established TSQ-SRM workflow used eight I364N clusterin glycopeptide glycoforms as precursors and two glycan-specific fragment oxonium ions at m/z 366.14 and m/z 657.24 as transitions (see Table 4). In addition, the peptide ion at m/z 574.56 representing [HN*STGCLR]2+ (SEQ ID NO: 2) where N* = HexNac asparagine residue, was included to serve as the third transition ion providing the confirmation of site-specific information. Details of each monitored transition are provided in Table 4.

T l 4. M l - RM n li i T .

In previous studies (data not shown), clusterin glycopeptides could be extracted from human serum without prior immunoprecipitation. Given the potential gains in sensitivity offered by SRM methods, a more direct geLC method for clusterin enrichment was followed that would be more compatible with high-throughput analysis, such as would be required for clinical diagnosis.

Initially, to identify the location of clusterin in a one-dimensional SDS-polyacrylamide gel electrophoresis (SDS-PAGE) experiment, normal human plasma (Dade-Behring, Germany) was depleted of albumin and IgG and extracted proteins. in Laemmli buffer and subjected to SDS-PAGE. Figure 13 shows the Gel-10 analysis of plasma albumin/IgG depletion. All ten bands were excised, reduced, alkylated, and trypsin digested prior to MS analysis on an Orbitrap Velos Pro. Clusterin glycopeptides were identified in the tryptic digests of bands #5, #6, and #7 (data not shown). most in band #7 with 45% peptide coverage.

The ten tryptic-digested gel bands were also submitted for analysis using the newly developed clusterin glycoform SRM method to confirm the suitability of the geLC method for sample preparation. Figure 14 shows that clusterin glycopeptides were found in bands #5, #6 and #7 and the majority of clusterin was identified in band #7, suggesting that these glyco-SRM results agree with the data from previous discoveries of Orbitrap Velos Pro.

When the SRM data files were examined for the extracted ion chromatograms (XIC) of eight I364N glycopeptide precursors (FIG. 15), it could be determined that most of them eluted between 7 and 8 minutes. Precise identification of elution time allows subsequent programming of SRM or adjustment of buffers. elution to improve the separation of closely related species, if more complex methods must be developed later.

It is a particular advantage of the present method that the integrated peak area for each monitored species can be used for label-free quantification using a summation scale approach. Furthermore, since this Gel10-glyco-SRM method does not require immunoprecipitation and clusterin glycopeptides can be detected in 30 minutes by TSQ, rather than one hour in Orbitrap, this new method provides a more efficient and faster way to verify the biomarker. clusterin glycopeptide potential. The same method can also be used for clinical evaluation of patient samples to aid in the diagnosis of MCI and Alzheimer's disease, as well as to provide prognostic information on the rate of hippocampal atrophy and cognitive decline. It would also be understood that the same SRM method can be applied with little further optimization to digested human plasma, serum, saliva, urine, or cerebrospinal fluid without the need for prior SDS-PAGE separation.

It is also possible to employ the same SRM method for the analysis of enriched clusterin from human plasma by immunoprecipitation. Therefore, to further validate our glycopeptide-specific biomarker, the glycol-SRM clusterin method was applied to assess immunoprecipitated clusterin from the Discovery Cohort. As expected, the SRM method yielded tighter quantitative results and this improved precision resulted in higher levels of significance for the reduction of specific glycoforms in Alzheimer patients with higher levels of hippocampal atrophy (Table 5). In total, five of the eight glycopeptides monitored in I364N were significantly reduced in cases of high atrophy.

A selection of box plots for SRM quantification of individual clusterin glycoforms is provided in Figure 12 .

Table 5 Significant changes in clusterin glycopeptides using the Gel10-glyco-SRM method (4 vs 4) discovery cohort

Example 7 - GEL10-glyco-SRM Selected Reaction Monitoring Assay Validation Study

The eighth GEL10-glyco-SRM clusterin glycopeptide assay developed in Example 6 was applied to the Validation Cohort analysis of plasma samples from patients with Alzheimer's disease, comprising 9 cases with [high] level of atrophy hippocampus and 10 cases with [low] level of hippocampal atrophy. Samples were as described in Table 2 and all sample preparation and analytical methods were performed as described in Example 6.

In this cohort, three I364N site-specific glycoforms showed a statistically significant difference between patients with high levels of hippocampal atrophy and those with lower rates of hippocampal atrophy (Table 6).

Ta l - R n imi n l n l1 - l - RM li f rm l rin n l h r v li ion

When the results for the Discovery and Validation Cohorts were combined, surprisingly, all eight glycoforms obtained statistical significance for reduced concentrations in the high atrophy group, compared to those with low hippocampal atrophy (Table 7). The ability of these eight clusterin glycopeptides to differentiate patients based on their hippocampal volume provides a minimally invasive means of diagnosing and predicting the progression of Alzheimer's disease.

Table 7 - Combined performance of Gel10-glyco-SRM from clusterin SRM glycoform in the combined discovery validation cohorts

Conclusions

A high-sensitivity SRM method for eight specific N-linked glycopeptides in human clusterin I364n can differentiate between cases of Alzheimer's disease with high hippocampal atrophy and cases of mild cognitive impairment with low hippocampal atrophy. This method may provide the basis for a routine clinical test to assess hippocampal atrophy, based on detecting the level of specific glycoforms in an individual patient and comparing it to levels known to represent specific levels of hippocampal atrophy.

References

Andreasen, N., C. Hesse, et al. (1999). "Cerebrospinal fluid beta-amyloid(1-42) in Alzheimer disease: differences between early- and late-onset Alzheimer disease and stability during the course of disease." Arch Neurol 56(6): 673-80.

Bosman, G.J., I.G. Bartholomeus, et al. (1991). "Erythrocyte membrane characteristics indicate abnormal cellular aging in patients with Alzheimer's disease." Neurobiol Aging 12(1): 13-8.

Butterfield, D.A., J.B. Owen (2011). "Lectin-affinity chromatography brain glycoproteomics and Alzheimer disease: insights into protein alterations consistent with the pathology and progression of this dementing disorder." Proteomics Clin Appl. 5(1-2) :50-6

Friedland, R.P. (1993). "Epidemiology, education and the ecology of Alzheimer's disease." Neurology 43(2): 246-9.

Ida, N., T. Hartmann, et al. (nineteen ninety six). "Analysis of heterogeneous A4 peptides in human cerebrospinal fluid and blood by a newly developed sensitive Western blot assay." J Biol Chem 271(37): 22908-14.

Kanai, M., E. Matsubara, et al. (1998). "Longitudinal study of cerebrospinal fluid levels of tau, A beta1-40 and A beta1-42(43) in Alzheimer's disease: a study in Japan." Ann Neurol 44(1): 17-26.

Kawarabayashi, T., L.H. Younkin, et al. (2001). "Age-dependent changes in brain, CSF and plasma amyloid (beta) protein in the Tg2576 transgenic mouse model of Alzheimer's disease." J Neurosci 21(2): 372-81.

Killick, R., E.M. Ribe, et al. (2012). "Clusterin regulates p-amyloid toxicity via Dickkopf-1-driven induction of the wnt-PCP-JNK pathway." Mol Psychiatry. doi: 10,1038/mp,2012,163. [Epub before print]

Kosaka, T., M. Imagawa, et al. (1997). "The beta APP717 Alzheimer mutation increases the percentage of plasma amyloidbeta protein ending at A beta42(43)." Neurology 48(3): 741-5.

Kuo, Y.M., T.A. Kokjohn, et al. (2000). "Elevated abeta42 in skeletal muscle of Alzheimer disease patients suggests peripheral alterations of AbetaPP metabolism." Am J Pathol 156(3): 797-805.

Lindner, M.D., D.D. Gordon, et al. (1993). "Increased levels of truncated nerve growth factor receptor in urine of mildly demented patients with Alzheimer's disease." Arch Neurol 50(10): 1054-60.

Nuutinen T., T. Suuronen, et al. (2009) "Clusterin: a forgotten player in Alzheimer's disease." Brain Res Rev. 61(2):89-104.

Claims (9)

1. A method for evaluating the level of hippocampal atrophy in a test subject with Alzheimer's disease (AD), characterized in that it comprises: (i) providing a protein-containing sample that has been obtained from the test subject; (ii) determining the concentration, amount or degree of expression of at least one glycoform of clusterin, where tetra-antennary glycoforms of the at least one glycoform of clusterin, as a proportion of the total of all glycoforms of the same glycoprotein fragment, they are quantified; (iii) comparing said concentration, amount or degree determined in (ii) with a reference of a control subject having AD with a low level of hippocampal atrophy; (iv) Based on the level of at least one glycoform of clusterin in the test subject relative to control, making an assessment of the level of hippocampal atrophy in the test subject, wherein a lower relative level of tetra-antennary glycoforms in the test subject sample, as compared to the relative level of tetra-antennary glycoforms in the control subject reference, indicates that the test subject is predicted to have a relatively higher level of hippocampal atrophy, wherein said at least one glycoform of clusterin comprises a glycosylated fragment of human clusterin consisting of HN*STGCLR (SEQ ID NO: 2), wherein "N*" indicates the glycan attachment residue, and wherein the protein-containing sample is selected from the group consisting of: blood plasma, blood cells, or serum. 2. The method according to claim 1, characterized in that said at least one glycoform of clusterin further comprises: a) any of the clusterin glycopeptides set forth in Table 3A, Table 3B, Table 3C, Table 5, Table 6 and/or Table 7; me b) an I364N glycan selected from the group consisting of: IJ64N_SA1-(HexNAc-Hex)2-central; IJ64N_SA2-(HexNAc-Hex)2-core; IJ64N_SA1-(HexNAc-Hex)3-core; IJ64N_SA2-(HexNAc-Hex)3-core; IJ64N_SA1-(HexNAc-Hex)4-core; IJ64N_SA3-(HexNAc-Hex)3-core; IJ64N_SA2-(HexNAc-Hex)4-middle; and IJ64N_SA3-(HexNAc-Hex)4-core. 3. The method according to claim 1 or claim 2, characterized in that the clusterin glycoforms are measured using: a) gel electrophoresis; either b) LC-MS/MS; and wherein, preferably, the relative amount of each glycoform is calculated in comparison to an equivalent heavy isotope labeled with a reference glycoform, using Selected Reaction Monitoring mass spectrometry. 4. The method according to any of claims 1 to 3, characterized in that the clusterin glycoforms are: a) measurements using summation-scale Selected Reaction Monitoring (SRM) mass spectrometry; and/or b) are not marked; 5. The method according to claim 3, characterized in that the heavy isotope labeled as reference clusterin glycoform is: a) a synthetic glycopeptide, wherein one or more heavy isotopes of H, C, N or O are substituted within the peptide or sugar components of said glycoform; or b) a natural, enriched clusterin glycoform that has been labeled with an isotopic mass tag, wherein said isotopic mass tag comprises one or more heavy isotopes of H, C, N, or O and wherein, said mass tag is capable of reacting with the peptide or sugar components of said glycoform. 6. The method according to claim 3, characterized in that the relative amount of each clusterin glycoform is calculated in comparison with an equivalent clusterin glycoform, labeled with an isobaric mass label, where: (i) each sample taken from the test subject is labeled with a member of an established isobaric mass label, to create a labeled analytical sample; (ii) a standard reference panel of enriched clusterin glycoforms is divided into two to six aliquots and each aliquot is labeled separately with additional members of the same set of isobaric mass labels as the labeled analytical sample and each aliquot labeled independently of the panel reference mixture is mixed in a predetermined ratio to create a clinically relevant concentration curve, as a standard reference mixture; (iii) an equal volume of the labeled analytical sample and the standard reference mixture are mixed, to form the MS calibrator sample; Y (iv) the MS calibrator sample prepared in step (iii) is analyzed by mass spectrometry; wherein, preferably, the isobaric mass tag set is a Tandem Mass Tag set. 7. The method according to any of the preceding claims, characterized in that the at least one glycoform of clusterin is measured by an immunological assay, wherein the immunological assay preferably comprises Western blot or ELISA. 8. A method for stratifying a plurality of test subjects according to their stage of Alzheimer's disease (AD), characterized in that it comprises: performing the method according to any one of claims 1 to 7, on at least one test sample from each of the human test subjects; Y based on the level of at least one specific clusterin glycoform in each of the test subjects, classifying the test subjects into more or less advanced stages of AD; wherein test subjects are classified according to their predicted degree of hippocampal atrophy. 9. A method for determining the efficacy of a treatment for Alzheimer's disease (AD) in a mammalian subject, comprising: determine the level of one or more glycoforms of clusterin by performing the method according to any of claims 1 to 7, prior to treatment of the subject and at least once during or after treatment of the subject, wherein the success of the treatment is demonstrated by the level of said or more clusterin glycoforms remaining stable or returning to more normal levels, wherein, the subject is human.